IEICE TRANSACTIONS on Information
Training of CNN with Heterogeneous Learning for Multiple Pedestrian Attributes Recognition Using Rarity Rate
Summary :
Pedestrian attribute information is important function for an advanced driver assistance system (ADAS). Pedestrian attributes such as body pose, face orientation and open umbrella indicate the intended action or state of the pedestrian. Generally, this information is recognized using independent classifiers for each task. Performing all of these separate tasks is too time-consuming at the testing stage. In addition, the processing time increases with increasing number of tasks. To address this problem, multi-task learning or heterogeneous learning is performed to train a single classifier to perform multiple tasks. In particular, heterogeneous learning is able to simultaneously train a classifier to perform regression and recognition tasks, which reduces both training and testing time. However, heterogeneous learning tends to result in a lower accuracy rate for classes with few training samples. In this paper, we propose a method to improve the performance of heterogeneous learning for such classes. We introduce a rarity rate based on the importance and class probability of each task. The appropriate rarity rate is assigned to each training sample. Thus, the samples in a mini-batch for training a deep convolutional neural network are augmented according to this rarity rate to focus on the classes with a few samples. Our heterogeneous learning approach with the rarity rate performs pedestrian attribute recognition better, especially for classes representing few training samples.
- Publication
- IEICE TRANSACTIONS on Information Vol.E101-D No.5 pp.1222-1231
- Publication Date
- 2018/05/01
- Publicized
- 2018/02/16
- Online ISSN
- 1745-1361
- DOI
- 10.1587/transinf.2017MVP0001
- Type of Manuscript
- Special Section PAPER (Special Section on Machine Vision and its Applications)
- Category
- Machine Vision and its Applications
Authors
Hiroshi FUKUI
Chubu University
Takayoshi YAMASHITA
Chubu University
Yuji YAMAUCHI
Chubu University
Hironobu FUJIYOSHI
Chubu University
Hiroshi MURASE
Nagoya University
Keyword
Latest Issue
Copyrights notice of machine-translated contents
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Cite this
Copy
Hiroshi FUKUI, Takayoshi YAMASHITA, Yuji YAMAUCHI, Hironobu FUJIYOSHI, Hiroshi MURASE, "Training of CNN with Heterogeneous Learning for Multiple Pedestrian Attributes Recognition Using Rarity Rate" in IEICE TRANSACTIONS on Information,
vol. E101-D, no. 5, pp. 1222-1231, May 2018, doi: 10.1587/transinf.2017MVP0001.
Abstract: Pedestrian attribute information is important function for an advanced driver assistance system (ADAS). Pedestrian attributes such as body pose, face orientation and open umbrella indicate the intended action or state of the pedestrian. Generally, this information is recognized using independent classifiers for each task. Performing all of these separate tasks is too time-consuming at the testing stage. In addition, the processing time increases with increasing number of tasks. To address this problem, multi-task learning or heterogeneous learning is performed to train a single classifier to perform multiple tasks. In particular, heterogeneous learning is able to simultaneously train a classifier to perform regression and recognition tasks, which reduces both training and testing time. However, heterogeneous learning tends to result in a lower accuracy rate for classes with few training samples. In this paper, we propose a method to improve the performance of heterogeneous learning for such classes. We introduce a rarity rate based on the importance and class probability of each task. The appropriate rarity rate is assigned to each training sample. Thus, the samples in a mini-batch for training a deep convolutional neural network are augmented according to this rarity rate to focus on the classes with a few samples. Our heterogeneous learning approach with the rarity rate performs pedestrian attribute recognition better, especially for classes representing few training samples.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2017MVP0001/_p
Copy
@ARTICLE{e101-d_5_1222,
author={Hiroshi FUKUI, Takayoshi YAMASHITA, Yuji YAMAUCHI, Hironobu FUJIYOSHI, Hiroshi MURASE, },
journal={IEICE TRANSACTIONS on Information},
title={Training of CNN with Heterogeneous Learning for Multiple Pedestrian Attributes Recognition Using Rarity Rate},
year={2018},
volume={E101-D},
number={5},
pages={1222-1231},
abstract={Pedestrian attribute information is important function for an advanced driver assistance system (ADAS). Pedestrian attributes such as body pose, face orientation and open umbrella indicate the intended action or state of the pedestrian. Generally, this information is recognized using independent classifiers for each task. Performing all of these separate tasks is too time-consuming at the testing stage. In addition, the processing time increases with increasing number of tasks. To address this problem, multi-task learning or heterogeneous learning is performed to train a single classifier to perform multiple tasks. In particular, heterogeneous learning is able to simultaneously train a classifier to perform regression and recognition tasks, which reduces both training and testing time. However, heterogeneous learning tends to result in a lower accuracy rate for classes with few training samples. In this paper, we propose a method to improve the performance of heterogeneous learning for such classes. We introduce a rarity rate based on the importance and class probability of each task. The appropriate rarity rate is assigned to each training sample. Thus, the samples in a mini-batch for training a deep convolutional neural network are augmented according to this rarity rate to focus on the classes with a few samples. Our heterogeneous learning approach with the rarity rate performs pedestrian attribute recognition better, especially for classes representing few training samples.},
keywords={},
doi={10.1587/transinf.2017MVP0001},
ISSN={1745-1361},
month={May},}
Copy
TY - JOUR
TI - Training of CNN with Heterogeneous Learning for Multiple Pedestrian Attributes Recognition Using Rarity Rate
T2 - IEICE TRANSACTIONS on Information
SP - 1222
EP - 1231
AU - Hiroshi FUKUI
AU - Takayoshi YAMASHITA
AU - Yuji YAMAUCHI
AU - Hironobu FUJIYOSHI
AU - Hiroshi MURASE
PY - 2018
DO - 10.1587/transinf.2017MVP0001
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E101-D
IS - 5
JA - IEICE TRANSACTIONS on Information
Y1 - May 2018
AB - Pedestrian attribute information is important function for an advanced driver assistance system (ADAS). Pedestrian attributes such as body pose, face orientation and open umbrella indicate the intended action or state of the pedestrian. Generally, this information is recognized using independent classifiers for each task. Performing all of these separate tasks is too time-consuming at the testing stage. In addition, the processing time increases with increasing number of tasks. To address this problem, multi-task learning or heterogeneous learning is performed to train a single classifier to perform multiple tasks. In particular, heterogeneous learning is able to simultaneously train a classifier to perform regression and recognition tasks, which reduces both training and testing time. However, heterogeneous learning tends to result in a lower accuracy rate for classes with few training samples. In this paper, we propose a method to improve the performance of heterogeneous learning for such classes. We introduce a rarity rate based on the importance and class probability of each task. The appropriate rarity rate is assigned to each training sample. Thus, the samples in a mini-batch for training a deep convolutional neural network are augmented according to this rarity rate to focus on the classes with a few samples. Our heterogeneous learning approach with the rarity rate performs pedestrian attribute recognition better, especially for classes representing few training samples.
ER -
English
